基于单目视觉的目标物高度测量

Single-view based height estimation of object

为简化现有单目测量方法中的标定前提条件和标定过程,保证测量的准确度,提出一种结合射影几何知识的自动化标定方法,完成目标物的高度测量,该标定法不用获得相机的内参和外参,只需获得固定相机拍摄图片中的灭点信息即可完成距离测量。在图片中自动找出标定物,对标定物进行一系列图像处理获取特征点,根据拟定的目标函数,采用遗传算法进行参数优化获得最优解,在遗传算法中采用柯西变异算子以获得较好的实验结果。实例验证了该方法在室内场景中的有效性。

To simplify the prerequisite and the process of calibration of the existing monocular measurement and to ensure the accuracy of the measurement,an automatic calibration method of height estimation based on projective geometry was put forward.Instead of calibrating intrinsic parameters,the approach aims to estimate the vanishing points of a stationary camera.The calibration process was formulated as finding the calibration parts,obtaining the feature points through the calibrated parts processing,and then defining a novel objective function to deal with the parameter optimization,in which twelve parameters for estimating vanishing points were defined.The genetic algorithm with Cauchy mutation operator was further used for obtaining robust results.Experimental results show the effectiveness of the proposed approach in interior scenes.